0
1

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?

st ocr clipboard

Last updated at Posted at 2024-06-25

import streamlit as st
from PIL import ImageGrab
import io

def get_image_from_clipboard():
    image = ImageGrab.grabclipboard()
    if image is not None:
        return image
    return None

def main():
    st.title("クリップボード画像アプリ")

    st.write("クリップボードに画像をコピーし、以下のボタンをクリックしてください。")

    if st.button("クリップボードから画像を貼り付け"):
        image = get_image_from_clipboard()
        if image is not None:
            st.image(image, caption="クリップボードからの画像", use_column_width=True)
        else:
            st.error("クリップボードに画像が見つかりませんでした。")

if __name__ == "__main__":
    main()



import streamlit as st
from PIL import ImageGrab, Image
import pandas as pd
import easyocr
import io
import numpy as np
from typing import List, Tuple

def get_image_from_clipboard():
    image = ImageGrab.grabclipboard()
    if image is not None:
        return image
    return None

def pil_to_bytes(image):
    img_byte_arr = io.BytesIO()
    image.save(img_byte_arr, format='PNG')
    img_byte_arr = img_byte_arr.getvalue()
    return img_byte_arr

def sort_by_y(element):
    return element[0][0][1]

def sort_by_x(element):
    return element[0][0][0]

def estimate_grid(elements: List[Tuple]) -> Tuple[List[float], List[float]]:
    x_coords = [bbox[0][0] for bbox, _ in elements]
    y_coords = [bbox[0][1] for bbox, _ in elements]
    
    x_coords.sort()
    y_coords.sort()
    
    def find_gaps(coords):
        gaps = []
        for i in range(1, len(coords)):
            gap = coords[i] - coords[i-1]
            if gap > 5:  # 5ピクセル以上の間隔を意味のあるギャップとする
                gaps.append((coords[i-1] + gap/2, gap))
        return sorted(gaps, key=lambda x: -x[1])[:10]  # 上位10個の大きなギャップを選択
    
    x_gaps = find_gaps(x_coords)
    y_gaps = find_gaps(y_coords)
    
    x_lines = [gap[0] for gap in x_gaps]
    y_lines = [gap[0] for gap in y_gaps]
    
    return x_lines, y_lines

def assign_to_cell(bbox, x_lines, y_lines):
    x = bbox[0][0]
    y = bbox[0][1]
    row = sum(1 for line in y_lines if line < y)
    col = sum(1 for line in x_lines if line < x)
    return row, col

def extract_table_structure(result):
    elements = [(bbox, text) for (bbox, text, prob) in result]
    elements.sort(key=sort_by_y)
    
    x_lines, y_lines = estimate_grid(elements)
    
    grid = {}
    for bbox, text in elements:
        row, col = assign_to_cell(bbox, x_lines, y_lines)
        if (row, col) in grid:
            grid[(row, col)] += " " + text
        else:
            grid[(row, col)] = text
    
    max_row = max(row for row, _ in grid.keys()) + 1
    max_col = max(col for _, col in grid.keys()) + 1
    
    df_data = [[grid.get((row, col), "") for col in range(max_col)] for row in range(max_row)]
    
    return pd.DataFrame(df_data)

def ocr_image_to_df(image):
    image_bytes = pil_to_bytes(image)
    reader = easyocr.Reader(['en', 'ja'])  # 言語を指定(必要に応じて調整)
    result = reader.readtext(image_bytes)

    # 表構造を抽出してデータフレームを作成
    df = extract_table_structure(result)

    return df, result

def main():
    st.title("クリップボード画像アプリ")

    st.write("クリップボードに画像をコピーし、以下のボタンをクリックしてください。")

    if st.button("クリップボードから画像を貼り付け"):
        image = get_image_from_clipboard()
        if image is not None:
            st.image(image, caption="クリップボードからの画像", use_column_width=True)

            # OCRを実行し、データフレームに変換
            df, ocr_result = ocr_image_to_df(image)

            # OCR結果のテキストを表示
            extracted_text = "\n".join([text for _, text, _ in ocr_result])
            st.write("抽出されたテキスト:")
            st.code(extracted_text)

            # データフレームを表示
            st.write("抽出された表:")
            st.dataframe(df)


        else:
            st.error("クリップボードに画像が見つかりませんでした。")

if __name__ == "__main__":
    main()

0
1
0

Register as a new user and use Qiita more conveniently

  1. You get articles that match your needs
  2. You can efficiently read back useful information
  3. You can use dark theme
What you can do with signing up
0
1

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?